From ca85f200998f988fd630658e05b87e1ab530551c Mon Sep 17 00:00:00 2001
From: Brad Jorsch <bjorsch@wikimedia.org>
Date: Mon, 28 Aug 2017 10:32:49 -0400
Subject: [PATCH] Make mw.uri.encode 'WIKI' mode match core {{urlencode:}}

The core {{urlencode:}} parser function doesn't encode various
characters in WIKI mode that it does in other modes. mw.uri.encode
should match that.

Bug: T174239
Change-Id: I2be0811cf39c02c5c0ad3433e4b0ef9030350e24
---
 engines/LuaCommon/lualib/mw.uri.lua                 | 13 ++++++++++++-
 tests/phpunit/engines/LuaCommon/UriLibraryTests.lua | 12 ++++++------
 2 files changed, 18 insertions(+), 7 deletions(-)

diff --git a/engines/LuaCommon/lualib/mw.uri.lua b/engines/LuaCommon/lualib/mw.uri.lua
index 6d01b054..ff00eae8 100644
--- a/engines/LuaCommon/lualib/mw.uri.lua
+++ b/engines/LuaCommon/lualib/mw.uri.lua
@@ -35,6 +35,17 @@ local function rawencode( s, space )
 	return ret
 end
 
+local function wikiencode( s )
+	local ret = string.gsub( s, '([^a-zA-Z0-9!$()*,./:;@~_-])', function ( c )
+		if c == ' ' then
+			return '_'
+		else
+			return string.format( '%%%02X', string.byte( c, 1, 1 ) )
+		end
+	end );
+	return ret
+end
+
 local function rawdecode( s )
 	local ret = string.gsub( s, '%%(%x%x)', function ( hex )
 		return string.char( tonumber( hex, 16 ) )
@@ -51,7 +62,7 @@ function uri.encode( s, enctype )
 	elseif enctype == 'PATH' then
 		return rawencode( s, '%20' )
 	elseif enctype == 'WIKI' then
-		return rawencode( s, '_' )
+		return wikiencode( s )
 	else
 		error( "bad argument #2 to 'encode' (expected QUERY, PATH, or WIKI)", 2 )
 	end
diff --git a/tests/phpunit/engines/LuaCommon/UriLibraryTests.lua b/tests/phpunit/engines/LuaCommon/UriLibraryTests.lua
index c01fa2e2..7bf68c93 100644
--- a/tests/phpunit/engines/LuaCommon/UriLibraryTests.lua
+++ b/tests/phpunit/engines/LuaCommon/UriLibraryTests.lua
@@ -18,16 +18,16 @@ local tests = {
 	  expect = { '__foo+b%C3%A1r+%2B+baz__' }
 	},
 	{ name = 'uri.encode QUERY', func = mw.uri.encode,
-	  args = { '__foo b\195\161r + baz__', 'QUERY' },
-	  expect = { '__foo+b%C3%A1r+%2B+baz__' }
+	  args = { '__foo b\195\161r + /baz/__', 'QUERY' },
+	  expect = { '__foo+b%C3%A1r+%2B+%2Fbaz%2F__' }
 	},
 	{ name = 'uri.encode PATH', func = mw.uri.encode,
-	  args = { '__foo b\195\161r + baz__', 'PATH' },
-	  expect = { '__foo%20b%C3%A1r%20%2B%20baz__' }
+	  args = { '__foo b\195\161r + /baz/__', 'PATH' },
+	  expect = { '__foo%20b%C3%A1r%20%2B%20%2Fbaz%2F__' }
 	},
 	{ name = 'uri.encode WIKI', func = mw.uri.encode,
-	  args = { '__foo b\195\161r + baz__', 'WIKI' },
-	  expect = { '__foo_b%C3%A1r_%2B_baz__' }
+	  args = { '__foo b\195\161r + /baz/__', 'WIKI' },
+	  expect = { '__foo_b%C3%A1r_%2B_/baz/__' }
 	},
 
 	{ name = 'uri.decode', func = mw.uri.decode,